Google DeepMind has unveiled AlphaGenome Atlas, a database that predicts the effects of every possible single nucleotide variant in the human genome — which the company describes as a high-resolution map of human DNA.
From AlphaFold to the Full Genome
DeepMind’s science program has spent a decade moving down the stack of biology: AlphaFold cracked protein structure prediction, AlphaGenome began interpreting the regulatory code of the genome, and Atlas now scales that to exhaustive coverage — every single-letter change that can occur in human DNA, with a predicted effect for each.
For genetics researchers, the significance is straightforward: the vast majority of human variants are classified as “of unknown significance.” A precomputed predictive map converts that unknown into a testable hypothesis, with implications for rare-disease diagnosis, drug target discovery, and precision medicine.
The AI-for-Science Portfolio
The launch headlines a heavy September for DeepMind’s science work, alongside WeatherNext 3 (weather forecasting) and agentic video understanding in Gemini. Read the full announcement on the Google blog.